Abstract The investigation of linguistic phenomena in corpora of spontaneous speech is sometimes hindered by corpus size or by the complexity of the factors influencing their occurrence. Language Production Experiments (LPEs) can specifically elicit such phenomena and can therefore be used to build corpora that allow for their investigation. Yet experiments are a wide category that covers very different tasks, and there is little empirical research that compares speakers’ response behavior to different task types. In this paper, we compare the responses of a group of 22 speakers to a translation task and a completion task, both of which target the syntactic phenomena complementizer agreement (CA). The results indicate that both experimental methods offer legitimate ways to investigate the phenomenon with specific advantages and disadvantages. However, a comparison of results from both tasks allows for insights that a single task could not have provided.

中文翻译：

---

作为语料库构建工具的语言产生实验：补语一致性的对比研究

摘要 对自发语音语料库中语言现象的研究有时会受到语料库规模或影响其发生的因素的复杂性的阻碍。语言生产实验 (LPE) 可以专门引发此类现象，因此可用于构建允许进行调查的语料库。然而，实验是一个涵盖非常不同任务的广泛类别，并且很少有实证研究将说话者的反应行为与不同任务类型进行比较。在本文中，我们比较了一组 22 位说话者对翻译任务和完成任务的响应，这两个任务都针对句法现象补充协议 (CA)。结果表明，两种实验方法都提供了研究具有特定优点和缺点的现象的合法方法。但是，比较两个任务的结果可以提供单个任务无法提供的见解。